DSLR/Mirrorless Camera Artefact Summary
Introductory Comments
I have written various articles about the image artefacts introduced by the internal processing performed by various DSLR/Mirrorless cameras.
The artefacts include:
- Star damage caused by raw data filtering (a.k.a. hot pixel suppression a.k.a. spatial filtering)
- Concentric rings in images, caused by Nikon's badly designed lossy compression algorithm
- Concentric rings (or polygons) in images, caused by lens corrections
- Concentric rings in images, caused by other hardcoded image corrections
The summary is aimed mainly at experienced astro-photographers (and aspiring astro-photographers!) who wish to select a DSLR/Mirrorless camera to take on challenges such as capturing very faint structures like molecular clouds and the IFN (integrated flux nebulosity) whilst faithfully capturing accurate star colours so that advanced techniques such as Photometric Colour Calibration can be used successfully.
Unfortunately, for such challenging targets, the processing workflow of calibration, stacking, background-subtraction and intensity-stretching that pulls the faint structures out of the background noise will also reveal the faint artefacts that have been introduced into the raw data by any internal processing performed by the camera.
You can help!
If you have a camera not listed below that you suspect has concentric rings caused by in-camera raw-processing then follow my test protocol for diagnosing baked-in concentric rings. If you're unsure how to process the resulting raw files then I'm always happy to receive them and analyse them.
Manufacturer Generalisations
Before looking at each camera in detail, there are some generalisations that can be made of internal data processing, manufacturer by manufacturer.
Sony mirrorless cameras suffer from 3 main problems:
- Raw data filtering (star eater or star eater v2) that erodes stars or gives a green bias to star colours. These effects are not noticeable for longer focal length optics that oversample stars.
- Lens corrections that create coloured concentric polygons when the camera is used with CPU lenses. This does not occur with non-CPU lenses or telescopes.
- Full-frame sensors have a left/right sensor split and resulting horizontal artefacts on the left hand side that do not calibrate out. These vary quite a lot from copy to copy for the same model of camera.
Nikon cameras suffer from 4 main problems:
- Raw data filtering that gives a green or magenta bias to star colours. More recent cameras are much less affected because they use more sophisticated raw data filtering algorithms. These effects are not noticeable for longer focal length optics that oversample stars.
- Coloured concentric rings in images caused by a badly designed lossy compression algorithm. This is only a problem for older cameras which don't have the option of lossless compression. These rings only occur with vignetting optics - the worse the vignetting, the worse the rings. For Windows users I have written a repair tool to fix raw data affected by the lossy compression.
- Coloured concentric rings and "waves" in images caused by some kind of hardcoded image correction (possibly a colour shading correction). This affects images whether or not the optics are vignetting.
- Full-frame Sony sensors in Nikon cameras have a left/right sensor split causing artefacts that do not calibrate out. These vary quite a lot from copy to copy for the same model of camera.
Fujifilm cameras are more difficult to generalise but many models suffer from heavy raw data filtering that causes:
- Star erosion and colour biases.
- Coloured concentric rings in images. These rings only occur with vignetting optics - the worse the vignetting, the worse the rings.
Canon cameras in general do not suffer any obvious effects from internal data processing. An exception is the EOS R and EOS Ra.
Unfortunately, I don't have any information on Olympus, Pentax or Leica cameras.
Quick navigation links:
Sony cameras:
- SonyA6000/A6300/A6500 SonyA1 SonyA7/A7R/A7S SonyA7II SonyA7RII SonyA7SII/A7III/A7RIII SonyA7SIII SonyA7IV SonyA7RIV SonyA7RV SonyA7CR SonyA7CII SonyZVE1
- Nikon D3400 NikonD5000 NikonD5100 NikonD5200 NikonD5300 NikonD5500 NikonD5600 NikonD7000 NikonD7500 NikonD500 NikonD6 NikonD600 NikonD700 NikonD750 NikonD800 NikonD810 NikonD810A NikonD850 NikonZ5 NikonZ50 NikonZ6/Z7 NikonZ6II NikonZ6III NikonZ8 NikonZ9
- Fujifilm X-T2 FujifilmX-T3 FujifilmX-T4 FujifilmX-T20 FujifilmX-T100 FujifilmX-T200 FujifilmGFX50R/GFX50S FujifilmGFX100/GFX100S
Sony Cameras
Here are some comments on individual Sony cameras:
Sony A6000/A6300/A6500
- Bulb-mode raw data filtering (a.k.a star eater) that damages unsaturated, undersampled stars. It is not noticeable for longer focal length optics that oversample stars. Manual exposures up to 30sec are unaffected.
- Lens corrections that create coloured concentric polygons when the camera is used with CPU lenses.
- Raw data filtering on longer exposures (a.k.a star eater v2) that damages unsaturated, undersampled stars, turning them green. It is not noticeable for longer focal length optics that oversample stars.
- Potential split sensor artefact and resulting horizontal banding on left hand side. This can be mitigated by shooting with the back-of-camera histogram halfway across.
- With lens corrections switched off, no evidence of the usual coloured concentric polygons.
- No evidence of histogram gaps caused by raw channel scaling, so no coloured concentric circular rings.
- Bulb-mode raw data filtering (a.k.a star eater) that damages unsaturated, undersampled stars. It is not noticeable for longer focal length optics that oversample stars. Manual exposures up to 30sec are unaffected.
- Colour channel scaling leading to regularly spaced histogram gaps that can cause coloured concentric rings with vignetting optics. This can be mitigated by shooting with the back-of-camera histogram halfway across.
- Potential split sensor artefact and resulting horizontal banding on left hand side. This can be mitigated by shooting with the back-of-camera histogram halfway across.
- No information on potential lens corrections.
- Bulb-mode raw data filtering (a.k.a star eater) that damages unsaturated, undersampled stars. It is not noticeable for longer focal length optics that oversample stars. Manual exposures up to 30sec are unaffected.
- Potential split sensor artefact and resulting horizontal banding on left hand side. This can be mitigated by shooting with the back-of-camera histogram halfway across.
- Colour channel scaling leading to regularly spaced histogram gaps that can cause coloured concentric rings with vignetting optics. This can be mitigated by shooting with the back-of-camera histogram halfway across.
- Lens corrections that create coloured concentric polygons when the camera is used with CPU lenses and cannot be disabled. This does not occur with non-CPU lenses or telescopes.
- Raw data filtering (a.k.a star eater v2) that damages unsaturated, undersampled stars, turning them green. It affects exposures of 4 seconds and longer. It is not noticeable for longer focal length optics that oversample stars. This does not affect exposures made in "continuous" mode.
- Potential split sensor artefact and resulting horizontal banding on left hand side. This can be mitigated by shooting with the back-of-camera histogram halfway across.
- Colour channel scaling leading to regularly spaced histogram gaps that can cause coloured concentric rings with vignetting optics. This can be mitigated by shooting with the back-of-camera histogram halfway across.
- Lens corrections that create coloured concentric polygons when the camera is used with CPU lenses and cannot be disabled. This does not occur with non-CPU lenses or telescopes.
- Additional comments: LifePixel reports that images from a modified A7RII are contaminated by emissions from an internal IR LED.
- Raw data filtering (a.k.a star eater v2) that damages unsaturated, undersampled stars, turning them green. It affects exposures of 4 seconds and longer. It is not noticeable for longer focal length optics that oversample stars.
- Potential split sensor artefact and resulting horizontal banding on left hand side. This can be mitigated by shooting with the back-of-camera histogram halfway across.
- Colour channel scaling leading to regularly spaced histogram gaps that can cause coloured concentric rings with vignetting optics. This can be mitigated by shooting with the back-of-camera histogram halfway across.
- Lens corrections that create coloured concentric polygons when the camera is used with CPU lenses and cannot be disabled. This does not occur with non-CPU lenses or telescopes.
- Additional comments: LifePixel reports that images from some modified models are contaminated by emissions from an internal IR LED.
- Raw data filtering (a.k.a star eater v2) that damages unsaturated, undersampled stars, turning them green. It affects exposures of 4 seconds and longer. It is not noticeable for longer focal length optics that oversample stars.
- Potential split sensor artefact and resulting horizontal banding on left hand side. This can be mitigated by shooting with the back-of-camera histogram halfway across.
- No evidence of histogram gaps caused by raw channel scaling, so no coloured concentric circular rings.
- Potential lens corrections that create coloured concentric polygons when the camera is used with CPU lenses and cannot be disabled. This does not occur with non-CPU lenses or telescopes.
- Raw data filtering on longer exposures (a.k.a star eater v2) that damages unsaturated, undersampled stars, turning them green. It is not noticeable for longer focal length optics that oversample stars.
- Potential split sensor artefact and resulting horizontal banding on left hand side. This can be mitigated by shooting with the back-of-camera histogram halfway across.
- With lens corrections switched off, no evidence of the usual coloured concentric polygons.
- No evidence of histogram gaps caused by raw channel scaling, so no coloured concentric circular rings.
- Additional comments: LifePixel reports that images from modified models are contaminated by emissions from an internal IR LED but only when using electronic shutter.
- Raw data filtering (a.k.a star eater v2) that damages unsaturated, undersampled stars, turning them green. It affects exposures of 3.2 seconds and longer. It is not noticeable for longer focal length optics that oversample stars.
- Potential split sensor artefact and resulting horizontal banding on left hand side. This can be mitigated by shooting with the back-of-camera histogram halfway across.
- With lens corrections switched off, no evidence of the usual coloured concentric polygons.
- Histogram gaps caused by raw channel scaling which that can cause coloured concentric rings with vignetting optics. This can be mitigated by shooting with the back-of-camera histogram halfway across.
- Additional comments: LifePixel reports that images from modified models are contaminated by emissions from an internal IR LED but only when using electronic shutter.
- No star damage from raw data filtering (a.k.a "star eater") on longer exposures
- Potential split sensor artefact and resulting horizontal banding on left hand side. This can be mitigated by shooting with the back-of-camera histogram halfway across.
- With lens corrections switched off, no evidence of the usual coloured concentric polygons.
- No evidence of histogram gaps caused by raw channel scaling, so no coloured concentric circular rings.
- Additional comments: LifePixel reports that images from modified models are contaminated by emissions from an internal IR LED but only when using electronic shutter.
- No star damage from raw data filtering (a.k.a "star eater") on longer exposures
- Potential split sensor artefact and resulting horizontal banding on left hand side. This can be mitigated by shooting with the back-of-camera histogram halfway across.
- With lens corrections switched off, no evidence of the usual coloured concentric polygons.
- No histogram gaps caused by raw channel scaling, so no coloured concentric circular rings.
- Additional comments: LifePixel reports that images from modified models are contaminated by emissions from an internal IR LED.
- No star damage from raw data filtering (a.k.a "star eater") on longer exposures
- Potential split sensor artefact and resulting horizontal banding on left hand side. This can be mitigated by shooting with the back-of-camera histogram halfway across.
- No information on potential coloured concentric polygons caused by lens corrections.
- No histogram gaps caused by raw channel scaling, so no coloured concentric circular rings.
- Additional comments: LifePixel reports that images from modified models are contaminated by emissions from an internal IR LED.
- No star damage from raw data filtering (a.k.a "star eater") on longer exposures
- Potential split sensor artefact and resulting horizontal banding on left hand side. This can be mitigated by shooting with the back-of-camera histogram halfway across.
- With lens corrections switched off, no evidence of the usual coloured concentric polygons.
- No evidence of histogram gaps caused by raw channel scaling, so no coloured concentric circular rings.
- No bulb mode, so no possibility of taking exposures longer than 30sec.
Nikon Cameras
Here are some comments on individual Nikon cameras:
Nikon D3400
- Raw data filtering that damages unsaturated, undersampled stars, turning them green.
- Coloured concentric rings in images caused by a badly designed lossy compression algorithm. Rings are more likely with vignetting optics - the worse the vignetting, the worse the rings.
- No information on potential hardcoded image correction.
- Severe raw data filtering (identical to Sony star eater) that damages unsaturated, undersampled stars.
- Coloured concentric rings in images caused by a badly designed lossy compression algorithm. Rings are more likely with vignetting optics - the worse the vignetting, the worse the rings. No information on which compression table is used, so it's not clear how to mitigate the problem.
- No information on potential hardcoded image correction.
- Bias level is zero which causes black clipping of bias and dark frames.
- Firmware hack disables the raw data filtering and the zero bias level which causes black clipping.
- Raw data filtering (identical to Sony star eater) that damages unsaturated, undersampled stars.
- Coloured concentric rings in images caused by a badly designed lossy compression algorithm. Rings are more likely with vignetting optics - the worse the vignetting, the worse the rings. This can be prevented by keeping the main peak of the histogram in each colour channel below 973.
- Coloured concentric rings and "waves" in images caused by hardcoded image correction. This affects images whether or not the optics are vignetting and occurs with both lenses and telescopes.
- Bias level is zero which causes black clipping of bias and dark frames.
- Firmware hack disables the raw data filtering and the zero bias level which causes black clipping but it does not prevent the hardcoded image correction.
- Raw data filtering (identical to Sony star eater) that damages unsaturated, undersampled stars.
- Coloured concentric rings in images caused by a badly designed lossy compression algorithm. Rings are more likely with vignetting optics - the worse the vignetting, the worse the rings. This can be prevented by keeping the main peak of the histogram in each colour channel below 973.
- Coloured concentric rings and "waves" in images caused by hardcoded image correction. This affects images whether or not the optics have vignetting and occurs with both lenses and telescopes.
- Bias level is zero which causes black clipping of bias and dark frames.
- Raw data filtering that damages unsaturated, undersampled stars, turning them green.
- Coloured concentric rings in images caused by a badly designed lossy compression algorithm. Rings are more likely with vignetting optics -the worse the vignetting, the worse the rings. Using 14-bit shooting, this can be prevented by keeping the main peak of the histogram in each colour channel below 973 (non bias-subtracted).
- Coloured concentric rings and "waves" in images caused by hardcoded image correction for CPU lenses. This affects images whether or not the optics have vignetting but does not affect non-CPU lenses nor telescopes.
- Bias level is 600 but black clipping of long exposure dark frames can still occur.
- Raw data filtering that damages unsaturated, undersampled stars, turning them green.
- Coloured concentric rings in images caused by a badly designed lossy compression algorithm. Rings are more likely with vignetting optics -the worse the vignetting, the worse the rings. Using 14-bit shooting, this can be prevented by keeping the main peak of the histogram in each colour channel below 2125 (non bias-subtracted).
- Coloured concentric rings and "waves" in images caused by hardcoded image correction for CPU lenses, non-CPU lenses and telescopes. This affects images whether or not the optics are vignetting.
- Bias level is 600 but black clipping of long exposure dark frames can still occur.
- Raw data filtering that damages unsaturated, undersampled stars, turning them green.
- Coloured concentric rings in images caused by a badly designed lossy compression algorithm. Rings are more likely with vignetting optics -the worse the vignetting, the worse the rings. Using 14-bit shooting, this can be prevented by keeping the main peak of the histogram in each colour channel below 2125 (non bias-subtracted).
- Coloured concentric rings and "waves" in images caused by hardcoded image correction for CPU lenses, non-CPU lenses and telescopes. This affects images whether or not the optics are vignetting.
- Bias level is 600 but I have no information on whether or not black clipping of long exposure dark frames can occur.
- Mild raw data filtering that causes minimal damage to unsaturated, undersampled stars because all 3 colour channels are taken into account.
- Lossless compression is available so there is no need to suffer the coloured concentric rings from the lossy compression.
- No information on potential hardcoded image correction.
- Bias level is zero which causes black clipping of bias and dark frames.
- Raw data filtering that damages unsaturated, undersampled stars, turning them green. Less severe than the Nikon 5xxx series of cameras.
- Lossless compression is available so there is no need to suffer the coloured concentric rings from the lossy compression.
- No information on potential hardcoded image correction.
- Bias level is 400 but I have no information on whether or not black clipping of long exposure dark frames can occur.
- Raw data filtering appears to be mild (similar to the D850?) and should cause minimal damage to unsaturated, undersampled stars
- Lossless compression is available so there is no need to suffer the coloured concentric rings from the lossy compression.
- No information on potential hardcoded image correction.
- Bias level is 400 with no evidence of black clipping
- No information on potential raw data filtering.
- Lossless compression is available so there is no need to suffer the coloured concentric rings from the lossy compression.
- No hardcoded correction causing coloured concentric rings when used with a telescope or lens that the firmware does not recognise. No information on the behaviour with a lens recognised by the firmware.
- Slight artefact caused by the full-frame split sensor.
- Bias level is 1008 with no evidence of black clipping.
- No information on potential raw data filtering.
- Lossless compression is available so there is no need to suffer the coloured concentric rings from the lossy compression.
- Coloured concentric rings when used with a lens recognised by the camera firmware. No information on the behaviour with unrecognised lenses and telescopes
- Slight artefact caused by the full-frame split sensor.
- Bias level is zero which causes black clipping of bias and dark frames.
- Severe raw data filtering (identical to Sony star eater) that damages unsaturated, undersampled stars.
- Lossless compression is available so there is no need to suffer the coloured concentric rings from the lossy compression.
- No information on potential hardcoded image correction.
- Bias level is zero which causes black clipping of bias and dark frames.
- No information on the potential full-frame split sensor artefacts.
- Firmware hack disables the raw data filtering and the zero bias level which causes black clipping.
- Raw data filtering that damages unsaturated, undersampled stars, turning them magenta because blue and red pixels are less likely to be capped.
- Lossless compression is available so there is no need to suffer the coloured concentric rings from the lossy compression.
- Coloured concentric rings in images caused by hardcoded image correction for CPU lenses, non-CPU lenses and telescopes. This affects images whether or not the optics are vignetting.
- Full-frame split sensor artefacts
- Bias level is 600 but I have no information on whether or not black clipping of long exposure dark frames can occur.
- Mild raw data filtering that causes minimal damage to unsaturated, undersampled stars because all 3 colour channels are taken into account.
- Lossless compression is available so there is no need to suffer the coloured concentric rings from the lossy compression.
- No information on potential hardcoded image correction.
- No information on the potential full-frame split sensor artefacts.
- Bias level is zero which causes black clipping of bias and dark frames.
- Firmware hack disables the raw data filtering and the zero bias level which causes black clipping.
- Raw data filtering that damages unsaturated, undersampled stars, possibly turning stars slightly pink because green pixels are more likely to be capped.
- Lossless compression is available so there is no need to suffer the coloured concentric rings from the lossy compression.
- Coloured concentric rings and "waves" in images caused by hardcoded image correction for CPU lenses. This affects images whether or not the optics have vignetting but I have no information on whether or not it affects non-CPU lenses and telescopes.
- No information on the potential full-frame split sensor artefacts.
- Bias level is 600 but black clipping of long exposure dark frames can occur.
- Mild raw data filtering that causes minimal damage to unsaturated, undersampled stars because all 3 colour channels are taken into account.
- Lossless compression is available so there is no need to suffer the coloured concentric rings from the lossy compression.
- No information on potential hardcoded image correction.
- No information on the potential full-frame split sensor artefacts.
- Bias level is 600 but black clipping of long exposure dark frames can occur.
- Mild raw data filtering that causes minimal damage to unsaturated, undersampled stars.
- Lossless compression is available so there is no need to suffer the coloured concentric rings from the lossy compression.
- Coloured concentric rings and "waves" in images caused by hardcoded image correction for CPU lenses. This affects images whether or not the optics are vignetting but does not affect non-CPU lenses nor telescopes.
- Split sensor and weird artefacts just to the right of the split. This can be mitigated by shooting with the back-of-camera histogram halfway across.
- Bias level is 400 and there is no evidence of black clipping of long exposure dark frames.
- No information on potential raw data filtering.
- Lossless compression is available so there is no need to suffer the coloured concentric rings from the lossy compression.
- Coloured concentric rings in images caused by hardcoded image correction for CPU lenses, non-CPU lenses and telescopes. This affects images whether or not the optics are vignetting.
- No information on the potential full-frame split sensor artefacts.
- Bias level is 1008 and there is no evidence of black clipping of long exposure dark frames.
- No information on potential raw data filtering.
- Lossless compression is available so there is no need to suffer the coloured concentric rings from the lossy compression.
- Coloured concentric rings and "waves" in images caused by hardcoded image correction for CPU lenses, non-CPU lenses and telescopes. This affects images whether or not the optics are vignetting.
- Bias level is 1008 and there is no evidence of black clipping of long exposure dark frames.
- Mild raw data filtering that causes minimal damage to unsaturated, undersampled stars with no apparent colour shifts.
- Lossless compression is available so there is no need to suffer the coloured concentric rings from the lossy compression.
- Coloured concentric rings and "waves" in images caused by hardcoded image correction for CPU lenses, non-CPU lenses and telescopes. This affects images whether or not the optics are vignetting.
- Split sensor artefacts. This can be mitigated by shooting with the back-of-camera histogram halfway across.
- Bias level is 1008 and there is no evidence of black clipping of long exposure dark frames.
- Mild raw data filtering that causes minimal damage to unsaturated, undersampled stars with no apparent colour shifts.
- Lossless compression is available so there is no need to suffer the coloured concentric rings from the lossy compression.
- Coloured concentric rings and "waves" in all 3 colour channels caused by hardcoded image correction for CPU lenses, non-CPU lenses and telescopes. This affects images whether or not the optics are vignetting.
- Potential split sensor artefacts. This can be mitigated by shooting with the back-of-camera histogram halfway across.
- Bias level is 1008 and there is no evidence of black clipping of long exposure dark frames.
- Raw data filtering but no information on potential damage to stars
- Lossless compression is available so there is no need to suffer the coloured concentric rings from the lossy compression.
- Coloured concentric rings in images caused by hardcoded image correction for CPU lenses. This affects images whether or not the optics are vignetting. No information about whether or not the rings are present for non-CPU lenses and telescopes.
- Potential split sensor artefacts. This can be mitigated by shooting with the back-of-camera histogram halfway across.
- Bias level is 1008 and there is no evidence of black clipping of long exposure dark frames.
- No information on potential raw data filtering.
- Lossless compression is available so there is no need to suffer the coloured concentric rings from the lossy compression.
- Coloured concentric rings in images caused by hardcoded image correction for CPU lenses, non-CPU lenses and telescopes. This affects images whether or not the optics are vignetting.
- Potential split sensor artefacts. This can be mitigated by shooting with the back-of-camera histogram halfway across.
- Bias level is 1008 and there is no evidence of black clipping of long exposure dark frames.
- No information on potential raw data filtering.
- Lossless compression is available so there is no need to suffer the coloured concentric rings from the lossy compression.
- Coloured concentric rings and "waves" in images caused by hardcoded image correction for CPU lenses. This affects images whether or not the optics are vignetting but I have no information on whether or not it affects non-CPU lenses and telescopes.
- No information on the potential full-frame split sensor artefacts.
- Bias level is 1008 and there is no evidence of black clipping of long exposure dark frames.
Fujifilm Cameras
Here are some comments on individual Fujifilm cameras:
Fujifilm X-T2
- Obvious raw data filtering in dark frames causing unsaturated, undersampled stars to turn magenta
- Histogram steps at intervals of 64 which potentially cause coloured concentric rings in images taken with vignetting optics
- Bias level is 1024 and there is no evidence of black clipping of long exposure dark frames.
- Obvious raw data filtering in dark frames causing unsaturated, undersampled stars to turn magenta
- Histogram steps at intervals of 64 which potentially cause coloured concentric rings in images taken with vignetting optics
- Bias level is 1024 and there is no evidence of black clipping of long exposure dark frames.
- Obvious raw data filtering in dark frames but no information on what damage this might potentially cause to unsaturated, undersampled stars
- Histogram steps at intervals of 64, demonstrated to cause coloured concentric rings in images taken with vignetting optics.
- Bias level is 1024 and there is no evidence of black clipping of long exposure dark frames.
- Obvious raw data filtering in dark frames but no information on what damage this might potentially cause to unsaturated, undersampled stars
- Histogram steps at intervals of 64, demonstrated to cause coloured concentric rings in images taken with vignetting optics.
- Bias level is 1024 and there is no evidence of black clipping of long exposure dark frames.
This appears to be a well-behaved camera i.e.
- No raw data filtering in dark frames.
- No histogram steps
- Bias level is 1024 and there is no evidence of black clipping of long exposure dark frames.
- Additional comments: Despite its name, the X-T100 uses a standard Bayer matrix sensor, not an XTrans sensor.
This appears to be a well-behaved camera i.e.
- No raw data filtering in dark frames.
- No histogram steps
- Bias level is 1024 and there is no evidence of black clipping of long exposure dark frames.
- Additional comments: Despite its name, the X-T200 uses a standard Bayer matrix sensor, not an XTrans sensor.
- Obvious raw data filtering in dark frames causing unsaturated, undersampled stars to turn magenta
- No evidence that the raw data filtering leads to coloured concentric rings in images
- Bias level is 1024 (variable) and there is no evidence of black clipping of long exposure dark frames.
- Obvious raw data filtering in dark frames but no information on what damage this might potentially cause to unsaturated, undersampled stars
- Histogram steps at intervals of 256, demonstrated to cause coloured concentric rings in images taken with vignetting optics.
- Bias level is 256 and there is no evidence of black clipping of long exposure dark frames.
Canon Cameras
Here are some comments on individual Canon cameras:
Canon EOS R
- Strange blips (or notches) in the histogram of the red channel, causing concentric rings in images taken with vignetting optics. This is caused by the camera scaling the red channel by a multiplier very close to unity. Earlier advice was to use a high ISO but now I've seen them appear also at ISO 1600. A potential fix to the raw data exists, as described in this Cloudy Nights post.
- Similar to the EOS R, strange blips (or notches) in the histogram of the red channel, causing concentric rings in images taken with vignetting optics.
- Additional comments: The EOS Ra suffers from red star halos which can be offset from the stars depending on the optics being used.
- Tests on flat files at ISO 100 showed no concentric rings
- Additional comments: The EOS R5 user manual confirms that in electronic shutter mode the bit depth drops from 14 bits to 12 bits and shutter speeds slower than 0.5 sec. are not available
- No information on potential concentric rings in the red channel at low ISO
- Additional comments: Alan Dyer's R6 review mentions that the R6 suffers from amp glow and that in electronic shutter mode the maximum exposure time is 1/2 second. The EOS R6 user manual confirms that in electronic shutter mode the bit depth drops from 14 bits to 12 bits and shutter speeds slower than 0.5 sec. are not available
- Confirmed concentric rings in red channel at low ISOs. These appear to have a different cause (as yet unknown) to the rings on the EOS R/Ra. The rings also appear at high ISOs e.g ISO 3200
- Additional comments: In electronic shutter mode the bit depth drops from 14 bits to 12 bits. The EOS R8 manual fails to mention this.
Sigma Cameras
Here are some comments on individual Sigma cameras:
Sigma FP
- Gaps in histogram of red channel and spikes in histogram of blue channel leading to weird rectangular structures in the background. This is clear evidence of raw data manipulation but it is not known what effect this may have on stacked deep-sky images.
Conclusion
It is surprising and depressing how much internal processing is taking place in many DSLR/Mirrorless cameras and how it can affect deep-sky astrophotography. Hopefully this summary will help astrophotographers navigate the maze of issues and make an informed camera choice.
If you have a camera not listed above that you suspect has concentric rings caused by in-camera raw-processing then follow my test protocol for diagnosing baked-in concentric rings. If you're unsure how to process the resulting raw files then I'm always happy to receive them and analyse them.
Last updated by Mark Shelley: 20 Nov 2024 (astro@markshelley.co.uk)